Scholarlink

<p>Agriculture has been the backbone of India’s economy for centuries, providing livelihood for millions of people and stimulating economic growth. Yield assessment is essential to agricultural planning and management,&#xa0;including crop insurance for farmers. Traditional methods of crop yield assessment adopted in the Republic of India are based on conducting a significant number of crop cutting experiments (CCE), which are time-consuming, labor-intensive, and expensive. Regression models based on meteorological data, Earth remote sensing data, and machine learning (ML) methods can significantly reduce the number of required CCEs. This article describes estimating yields using regression models based on ML methods and ensembles. ML models based on the following methods were used: Deep Neural Networks (DNN), Random Forest, Extremely Randomized Trees, Catboost, K-Nearest Neighbors, and Linear Regression. Separate models and ensembles were built for each “season–district–crop” combination. Yield evaluation was conducted in 44 districts of Andhra Pradesh, Haryana, Jharkhand, Madhya Pradesh, Odisha, Tamil Nadu, Uttar Pradesh, Bihar, and Rajasthan. The best coefficient of determination reached 0.94. The minimum mean absolute percentage error was 6%. For most of the districts studied, the ensembling resulted in an increase of the <InlineEquation ID="IEq1"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="12524_2025_2186_Article_IEq1.gif" Format="GIF" Height="16" Rendition="HTML" Resolution="72" Type="Linedraw" Width="21" /> </InlineMediaObject> <EquationSource Format="TEX">\(R^2\)</EquationSource> <EquationSource Format="MATHML"><math> <msup> <mi>R</mi> <mn>2</mn> </msup> </math></EquationSource> </InlineEquation> metric by 2–3 % points.</p> Graphical Abstract <p></p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Ensemble Machine Learning Models for Rice and Wheat Yield Prediction: A Comparative Study Across Districts in India’s Kharif and Rabi Seasons

  • 期刊论文

  • 发表日期：

    2025年4月28日

查看全文
  • Vladimir Zhdanov,
  • Dmitry Kaplin,
  • Mikhail Akimov,
  • Alexey Antonov,
  • Mikhail Gasanov,
  • Vitaly Malchenko

摘要

Agriculture has been the backbone of India’s economy for centuries, providing livelihood for millions of people and stimulating economic growth. Yield assessment is essential to agricultural planning and management, including crop insurance for farmers. Traditional methods of crop yield assessment adopted in the Republic of India are based on conducting a significant number of crop cutting experiments (CCE), which are time-consuming, labor-intensive, and expensive. Regression models based on meteorological data, Earth remote sensing data, and machine learning (ML) methods can significantly reduce the number of required CCEs. This article describes estimating yields using regression models based on ML methods and ensembles. ML models based on the following methods were used: Deep Neural Networks (DNN), Random Forest, Extremely Randomized Trees, Catboost, K-Nearest Neighbors, and Linear Regression. Separate models and ensembles were built for each “season–district–crop” combination. Yield evaluation was conducted in 44 districts of Andhra Pradesh, Haryana, Jharkhand, Madhya Pradesh, Odisha, Tamil Nadu, Uttar Pradesh, Bihar, and Rajasthan. The best coefficient of determination reached 0.94. The minimum mean absolute percentage error was 6%. For most of the districts studied, the ensembling resulted in an increase of the \(R^2\) R 2 metric by 2–3 % points.

Graphical Abstract